 zrm
|
b853d45601
ggml : add NUMA support (#1556)
|
2 lat temu |
 Didzis Gosko
|
527b6fba1d
llama : make model stateless and context stateful (llama_state) (#1797)
|
2 lat temu |
 Johannes Gäßler
|
2c9380dd2f
Only one CUDA stream per device for async compute (#1898)
|
2 lat temu |
 Borislav Stanimirov
|
9cbf50c041
build : fix and ignore MSVC warnings (#1889)
|
2 lat temu |
|
Johannes Gäßler
|
6b8312e797
Better error when using both LoRA + GPU layers (#1861)
|
2 lat temu |
|
Johannes Gäßler
|
254a7a7a5f
CUDA full GPU acceleration, KV cache in VRAM (#1827)
|
2 lat temu |
 Kerfuffle
|
fa84c4b3e8
Fix issue where interactive mode crashes when input exceeds ctx size (#1789)
|
2 lat temu |
 Willy Tarreau
|
35a84916fb
main: add the possibility to open the prompt cache read-only (#1640)
|
2 lat temu |
|
Johannes Gäßler
|
17366df842
Multi GPU support, CUDA refactor, CUDA scratch buffer (#1703)
|
2 lat temu |
 Georgi Gerganov
|
ecb217db4f
llama : Metal inference (#1642)
|
2 lat temu |
|
Kerfuffle
|
1b78ed2081
Only show -ngl option when relevant + other doc/arg handling updates (#1625)
|
2 lat temu |
 Vladimir Zorin
|
337aea1139
examples : add --alias option to gpt_params to set use friendly model name (#1614)
|
2 lat temu |
 DannyDaemonic
|
d2c59b8ba4
Fix for mingw (#1462)
|
2 lat temu |
 Jason McCartney
|
7694b52b9a
main : make reverse prompt option act as a stop token in non-interactive mode (#1032)
|
2 lat temu |
|
Georgi Gerganov
|
4b7e245adf
minor : fix compile warnings
|
2 lat temu |
 Stephan Walter
|
dc271c52ed
Remove unused n_parts parameter (#1509)
|
2 lat temu |
|
zrm
|
63d20469b8
fix get_num_physical_cores() (#1436)
|
2 lat temu |
|
Johannes Gäßler
|
905d87b70a
ggml : GPU-accelerated token generation (#1412)
|
2 lat temu |
|
Johannes Gäßler
|
773ee249fb
CLI args use - instead of _, backwards compatible (#1416)
|
2 lat temu |
 Evan Jones
|
cf348a60e0
main : add option to save full output to session (#1338)
|
2 lat temu |
|
DannyDaemonic
|
e6a46b0ed1
Locale fix for Windows (#1379)
|
2 lat temu |
|
DannyDaemonic
|
41654efea8
Interface improvements and `--multiline-input` (previously `--author-mode`) (#1040)
|
2 lat temu |
|
Georgi Gerganov
|
f9a6364912
llama : require first token to be BOS (#1303)
|
2 lat temu |
|
Johannes Gäßler
|
1f48b0abcf
Documented CUDA reproducibility, added warning (#1346)
|
2 lat temu |
 44670
|
2edbdb0f99
main : add --in-suffix option (#1318)
|
2 lat temu |
|
DannyDaemonic
|
db1080876a
Only escape prompts when used with `-e` (#1311)
|
2 lat temu |
|
DannyDaemonic
|
2485d7a4d3
Process escape sequences given in prompts (#1173)
|
2 lat temu |
 slaren
|
bf4b22ffe4
fix missing parameters in `llama_init_from_gpt_params` (#1293)
|
2 lat temu |
 Ron Evans
|
67c77799e0
examples : add llama_init_from_gpt_params() common function (#1290)
|
2 lat temu |
 Robert Brisita
|
2bb992f034
llama : allow 0 as a seed number. (#1275)
|
2 lat temu |
